Until recently, most of what we knew about the emergence of autism in infancy relied on retrospective accounts. A new approach, the study of infants at familial risk, aims to identify the earliest “pure” manifestations of autism, before subsequent years of atypical development exacerbate, or compensate for, initial atypical development. An emerging picture from these studies is that early impairments in one or more of several functional cognitive systems are associated, respectively, with familial risk and with a later autism diagnosis. Understanding the temporal associations between these impairments over time will reveal the underlying mechanisms of atypical development in autism and inform approaches to early intervention. I will present new data from our ongoing prospective longitudinal and intervention studies.